Re: Relational Databases and Their Guts

From: Todd Bandrowsky <anakin_at_unitedsoftworks.com>
Date: 21 Jun 2003 07:02:24 -0700
Message-ID: <af3d9224.0306210602.294119ab_at_posting.google.com>


>
> Thanks for the clarification.

You are quite welcome.

>
> In this newsgroup, I think it would be clarifying if you called your first
> category SQL RDBMS.
>
> Fabain Pascal uses the term TRDBMS for "Truly Relational Database Managements
> System(s)". This term is OK when you want to be explicit about a RDBMS that is
> truer to the relational model, than SQL RDBMSes can ever be.

That's what I'm trying to bumble my way into getting at. So, SQL Server, Oracle, MySQL are NOT RDMBS, because even though they are DBMS, they are not relational. Now, let's see if I can get a better understanding of WHY SQL Server and Oracle are not true RDBMS.

Please correct me.

First off, the information in this link is wrong then, as it claims SQL implements Relational.

http://wombat.doc.ic.ac.uk/foldoc/foldoc.cgi?relational+data+model

But this is more accurate, as it defines the basic model of Relational Algebra.

http://www.wikipedia.org/wiki/Relational+algebra

And,

So called RDBMS are not relational because they do not implement some basic operators:

  1. do not implement subtraction R - S.
  2. the project operation is merged with the select operation into a single confusing statement.
  3. do not implement division either. What IS division in relational algebra?
  4. In relational algebra, according to Codd 70, primary keys are allowed to be redundant across a domain

Further

c) they aren't called tables, they are called relations, although I think Codd used the term domains. But he also used domains to describe related things within a relation, introducing that notion to define 1NF.

d) they aren't called rows, they are called n-tuples.

What is the relationship between predicate logic and relational calculus, and what did Codd conceptually do when he developed an algorithm to reduce relational calculus to relational algebra? Did this show that relational algebra can represent all knowledge representable by predicate logic?

Questions:

What does relational division do? I'm just lost on th8is one.

What is the ith thingy in an n tuple called? Received on Sat Jun 21 2003 - 16:02:24 CEST

Original text of this message